Parallel-vector computation with high-p element-by-element methods

Abstract

A basic element-by-element formulation of the finite element method is presented. This is coupled with a bi-conjugate gradient iterative solver with Jacobi and block type preconditioners. A variety of performance calculations for linear and nonlinear test problems are presented. Results include parallel-vector perfor-mance studies on the Cray YMP and the Alliant FX/8.

Keywords:

• p. finite element methods
• iterative methods
• parallel/vector supercomputing

C.R. CATEGORIES::

• G1.3. G1.8.